主催: Eastern Asia Society for Transportation Studies
p. 81
LISREL and neural networks have recently become popular methodology for causal models. However, few researches have applied either of these two methods in behavioral research of passengers. To investigate the differences between LISREL and neural networks in research of passengers' behavioral intentions, this study applied these two methods to a case of inter city bus transport in Taiwan. First, we applied LISREL to test the goodness of fit of the research model. Second, two competitive models of the NN model were tested, in which one was the full connected network and the other was the non-full connection network. The results indicated that LISREL can be a convenient and effective analysis tool if the causal relationships are known. At the same time, no matter if the causality has been derived in advance or not, NN provided a suitable prediction after the proper training procedure.
